« back to all changes in this revision
Viewing changes to bzrlib/diff.py
- browse files at revision 628
- compare with another revision
- download diff
- download tarball
- view history from revision 628
- Committer: Martin Pool
- Date: 2005-06-06 11:53:29 UTC
- Revision ID: mbp@sourcefrog.net-20050606115329-1596352add25bffd
- merge aaron's updated merge/pull code
- files added:
- .rsyncexclude
- contrib
- contrib/bash
- contrib/zsh
- elementtree
- patches
- urlgrabber
- files removed:
- bzrlib/tests.py
- files modified:
- .bzrignore
added
removed
bzrlib/diff.py
15
15
# along with this program; if not, write to the Free Software
16
16
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
17
18
from sets import Set
19
20
18
from trace import mutter
21
22
23
24
25
26
27
def diff_trees(old_tree, new_tree):
28
"""Compute diff between two trees.
29
30
They may be in different branches and may be working or historical
31
trees.
32
33
Yields a sequence of (state, id, old_name, new_name, kind).
34
Each filename and each id is listed only once.
35
"""
36
37
## TODO: Compare files before diffing; only mention those that have changed
38
39
## TODO: Set nice names in the headers, maybe include diffstat
40
41
## TODO: Perhaps make this a generator rather than using
42
## a callback object?
43
44
## TODO: Allow specifying a list of files to compare, rather than
45
## doing the whole tree? (Not urgent.)
46
47
## TODO: Allow diffing any two inventories, not just the
48
## current one against one. We mgiht need to specify two
49
## stores to look for the files if diffing two branches. That
50
## might imply this shouldn't be primarily a Branch method.
51
52
## XXX: This doesn't report on unknown files; that can be done
53
## from a separate method.
54
55
old_it = old_tree.list_files()
56
new_it = new_tree.list_files()
57
58
def next(it):
59
try:
60
return it.next()
61
except StopIteration:
62
return None
63
64
old_item = next(old_it)
65
new_item = next(new_it)
66
67
# We step through the two sorted iterators in parallel, trying to
68
# keep them lined up.
69
70
while (old_item != None) or (new_item != None):
71
# OK, we still have some remaining on both, but they may be
72
# out of step.
73
if old_item != None:
74
old_name, old_class, old_kind, old_id = old_item
75
else:
76
old_name = None
77
78
if new_item != None:
79
new_name, new_class, new_kind, new_id = new_item
80
else:
81
new_name = None
82
83
mutter(" diff pairwise %r" % (old_item,))
84
mutter(" %r" % (new_item,))
85
86
if old_item:
87
# can't handle the old tree being a WorkingTree
88
assert old_class == 'V'
89
90
if new_item and (new_class != 'V'):
91
yield new_class, None, None, new_name, new_kind
92
new_item = next(new_it)
93
elif (not new_item) or (old_item and (old_name < new_name)):
94
mutter(" extra entry in old-tree sequence")
95
if new_tree.has_id(old_id):
96
# will be mentioned as renamed under new name
97
pass
98
else:
99
yield 'D', old_id, old_name, None, old_kind
100
old_item = next(old_it)
101
elif (not old_item) or (new_item and (new_name < old_name)):
102
mutter(" extra entry in new-tree sequence")
103
if old_tree.has_id(new_id):
104
yield 'R', new_id, old_tree.id2path(new_id), new_name, new_kind
105
else:
106
yield 'A', new_id, None, new_name, new_kind
107
new_item = next(new_it)
108
elif old_id != new_id:
109
assert old_name == new_name
110
# both trees have a file of this name, but it is not the
111
# same file. in other words, the old filename has been
112
# overwritten by either a newly-added or a renamed file.
113
# (should we return something about the overwritten file?)
114
if old_tree.has_id(new_id):
115
# renaming, overlying a deleted file
116
yield 'R', new_id, old_tree.id2path(new_id), new_name, new_kind
117
else:
118
yield 'A', new_id, None, new_name, new_kind
119
120
new_item = next(new_it)
121
old_item = next(old_it)
122
else:
123
assert old_id == new_id
124
assert old_name == new_name
125
assert old_kind == new_kind
126
127
if old_kind == 'directory':
128
yield '.', new_id, old_name, new_name, new_kind
129
elif old_tree.get_file_size(old_id) != new_tree.get_file_size(old_id):
130
mutter(" file size has changed, must be different")
131
yield 'M', new_id, old_name, new_name, new_kind
132
elif old_tree.get_file_sha1(old_id) == new_tree.get_file_sha1(old_id):
133
mutter(" SHA1 indicates they're identical")
134
## assert compare_files(old_tree.get_file(i), new_tree.get_file(i))
135
yield '.', new_id, old_name, new_name, new_kind
136
else:
137
mutter(" quick compare shows different")
138
yield 'M', new_id, old_name, new_name, new_kind
139
140
new_item = next(new_it)
141
old_item = next(old_it)
142
143
19
from errors import BzrError
20
21
22
def internal_diff(old_label, oldlines, new_label, newlines, to_file):
23
import difflib
24
25
# FIXME: difflib is wrong if there is no trailing newline.
26
# The syntax used by patch seems to be "\ No newline at
27
# end of file" following the last diff line from that
28
# file. This is not trivial to insert into the
29
# unified_diff output and it might be better to just fix
30
# or replace that function.
31
32
# In the meantime we at least make sure the patch isn't
33
# mangled.
34
35
36
# Special workaround for Python2.3, where difflib fails if
37
# both sequences are empty.
38
if not oldlines and not newlines:
39
return
40
41
nonl = False
42
43
if oldlines and (oldlines[-1][-1] != '\n'):
44
oldlines[-1] += '\n'
45
nonl = True
46
if newlines and (newlines[-1][-1] != '\n'):
47
newlines[-1] += '\n'
48
nonl = True
49
50
ud = difflib.unified_diff(oldlines, newlines,
51
fromfile=old_label, tofile=new_label)
52
53
# work-around for difflib being too smart for its own good
54
# if /dev/null is "1,0", patch won't recognize it as /dev/null
55
if not oldlines:
56
ud = list(ud)
57
ud[2] = ud[2].replace('-1,0', '-0,0')
58
elif not newlines:
59
ud = list(ud)
60
ud[2] = ud[2].replace('+1,0', '+0,0')
61
62
to_file.writelines(ud)
63
if nonl:
64
print >>to_file, "\\ No newline at end of file"
65
print >>to_file
66
67
68
69
70
def external_diff(old_label, oldlines, new_label, newlines, to_file,
71
diff_opts):
72
"""Display a diff by calling out to the external diff program."""
73
import sys
74
75
if to_file != sys.stdout:
76
raise NotImplementedError("sorry, can't send external diff other than to stdout yet",
77
to_file)
78
79
# make sure our own output is properly ordered before the diff
80
to_file.flush()
81
82
from tempfile import NamedTemporaryFile
83
import os
84
85
oldtmpf = NamedTemporaryFile()
86
newtmpf = NamedTemporaryFile()
87
88
try:
89
# TODO: perhaps a special case for comparing to or from the empty
90
# sequence; can just use /dev/null on Unix
91
92
# TODO: if either of the files being compared already exists as a
93
# regular named file (e.g. in the working directory) then we can
94
# compare directly to that, rather than copying it.
95
96
oldtmpf.writelines(oldlines)
97
newtmpf.writelines(newlines)
98
99
oldtmpf.flush()
100
newtmpf.flush()
101
102
if not diff_opts:
103
diff_opts = []
104
diffcmd = ['diff',
105
'--label', old_label,
106
oldtmpf.name,
107
'--label', new_label,
108
newtmpf.name]
109
110
# diff only allows one style to be specified; they don't override.
111
# note that some of these take optargs, and the optargs can be
112
# directly appended to the options.
113
# this is only an approximate parser; it doesn't properly understand
114
# the grammar.
115
for s in ['-c', '-u', '-C', '-U',
116
'-e', '--ed',
117
'-q', '--brief',
118
'--normal',
119
'-n', '--rcs',
120
'-y', '--side-by-side',
121
'-D', '--ifdef']:
122
for j in diff_opts:
123
if j.startswith(s):
124
break
125
else:
126
continue
127
break
128
else:
129
diffcmd.append('-u')
130
131
if diff_opts:
132
diffcmd.extend(diff_opts)
133
134
rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
135
136
if rc != 0 and rc != 1:
137
# returns 1 if files differ; that's OK
138
if rc < 0:
139
msg = 'signal %d' % (-rc)
140
else:
141
msg = 'exit code %d' % rc
142
143
raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
144
finally:
145
oldtmpf.close() # and delete
146
newtmpf.close()
147
148
149
150
def show_diff(b, revision, specific_files, external_diff_options=None):
151
"""Shortcut for showing the diff to the working tree.
152
153
b
154
Branch.
155
156
revision
157
None for each, or otherwise the old revision to compare against.
158
159
The more general form is show_diff_trees(), where the caller
160
supplies any two trees.
161
"""
162
import sys
163
164
if revision == None:
165
old_tree = b.basis_tree()
166
else:
167
old_tree = b.revision_tree(b.lookup_revision(revision))
168
169
new_tree = b.working_tree()
170
171
show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
172
external_diff_options)
173
174
175
176
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
177
external_diff_options=None):
178
"""Show in text form the changes from one tree to another.
179
180
to_files
181
If set, include only changes to these files.
182
183
external_diff_options
184
If set, use an external GNU diff and pass these options.
185
"""
186
187
# TODO: Options to control putting on a prefix or suffix, perhaps as a format string
188
old_label = ''
189
new_label = ''
190
191
DEVNULL = '/dev/null'
192
# Windows users, don't panic about this filename -- it is a
193
# special signal to GNU patch that the file should be created or
194
# deleted respectively.
195
196
# TODO: Generation of pseudo-diffs for added/deleted files could
197
# be usefully made into a much faster special case.
198
199
if external_diff_options:
200
assert isinstance(external_diff_options, basestring)
201
opts = external_diff_options.split()
202
def diff_file(olab, olines, nlab, nlines, to_file):
203
external_diff(olab, olines, nlab, nlines, to_file, opts)
204
else:
205
diff_file = internal_diff
206
207
208
delta = compare_trees(old_tree, new_tree, want_unchanged=False,
209
specific_files=specific_files)
210
211
for path, file_id, kind in delta.removed:
212
print '*** removed %s %r' % (kind, path)
213
if kind == 'file':
214
diff_file(old_label + path,
215
old_tree.get_file(file_id).readlines(),
216
DEVNULL,
217
[],
218
to_file)
219
220
for path, file_id, kind in delta.added:
221
print '*** added %s %r' % (kind, path)
222
if kind == 'file':
223
diff_file(DEVNULL,
224
[],
225
new_label + path,
226
new_tree.get_file(file_id).readlines(),
227
to_file)
228
229
for old_path, new_path, file_id, kind, text_modified in delta.renamed:
230
print '*** renamed %s %r => %r' % (kind, old_path, new_path)
231
if text_modified:
232
diff_file(old_label + old_path,
233
old_tree.get_file(file_id).readlines(),
234
new_label + new_path,
235
new_tree.get_file(file_id).readlines(),
236
to_file)
237
238
for path, file_id, kind in delta.modified:
239
print '*** modified %s %r' % (kind, path)
240
if kind == 'file':
241
diff_file(old_label + path,
242
old_tree.get_file(file_id).readlines(),
243
new_label + path,
244
new_tree.get_file(file_id).readlines(),
245
to_file)
246
247
248
249
class TreeDelta(object):
250
"""Describes changes from one tree to another.
251
252
Contains four lists:
253
254
added
255
(path, id, kind)
256
removed
257
(path, id, kind)
258
renamed
259
(oldpath, newpath, id, kind, text_modified)
260
modified
261
(path, id, kind)
262
unchanged
263
(path, id, kind)
264
265
Each id is listed only once.
266
267
Files that are both modified and renamed are listed only in
268
renamed, with the text_modified flag true.
269
270
The lists are normally sorted when the delta is created.
271
"""
272
def __init__(self):
273
self.added = []
274
self.removed = []
275
self.renamed = []
276
self.modified = []
277
self.unchanged = []
278
279
def has_changed(self):
280
changes = len(self.added) + len(self.removed) + len(self.renamed)
281
changes += len(self.modified)
282
return (changes != 0)
283
284
def touches_file_id(self, file_id):
285
"""Return True if file_id is modified by this delta."""
286
for l in self.added, self.removed, self.modified:
287
for v in l:
288
if v[1] == file_id:
289
return True
290
for v in self.renamed:
291
if v[2] == file_id:
292
return True
293
return False
294
295
296
def show(self, to_file, show_ids=False, show_unchanged=False):
297
def show_list(files):
298
for path, fid, kind in files:
299
if kind == 'directory':
300
path += '/'
301
elif kind == 'symlink':
302
path += '@'
303
304
if show_ids:
305
print >>to_file, ' %-30s %s' % (path, fid)
306
else:
307
print >>to_file, ' ', path
308
309
if self.removed:
310
print >>to_file, 'removed:'
311
show_list(self.removed)
312
313
if self.added:
314
print >>to_file, 'added:'
315
show_list(self.added)
316
317
if self.renamed:
318
print >>to_file, 'renamed:'
319
for oldpath, newpath, fid, kind, text_modified in self.renamed:
320
if show_ids:
321
print >>to_file, ' %s => %s %s' % (oldpath, newpath, fid)
322
else:
323
print >>to_file, ' %s => %s' % (oldpath, newpath)
324
325
if self.modified:
326
print >>to_file, 'modified:'
327
show_list(self.modified)
328
329
if show_unchanged and self.unchanged:
330
print >>to_file, 'unchanged:'
331
show_list(self.unchanged)
332
333
334
335
def compare_trees(old_tree, new_tree, want_unchanged, specific_files=None):
336
"""Describe changes from one tree to another.
337
338
Returns a TreeDelta with details of added, modified, renamed, and
339
deleted entries.
340
341
The root entry is specifically exempt.
342
343
This only considers versioned files.
344
345
want_unchanged
346
If true, also list files unchanged from one version to
347
the next.
348
349
specific_files
350
If true, only check for changes to specified names or
351
files within them.
352
"""
353
354
from osutils import is_inside_any
355
356
old_inv = old_tree.inventory
357
new_inv = new_tree.inventory
358
delta = TreeDelta()
359
mutter('start compare_trees')
360
361
# TODO: match for specific files can be rather smarter by finding
362
# the IDs of those files up front and then considering only that.
363
364
for file_id in old_tree:
365
if file_id in new_tree:
366
kind = old_inv.get_file_kind(file_id)
367
assert kind == new_inv.get_file_kind(file_id)
368
369
assert kind in ('file', 'directory', 'symlink', 'root_directory'), \
370
'invalid file kind %r' % kind
371
372
if kind == 'root_directory':
373
continue
374
375
old_path = old_inv.id2path(file_id)
376
new_path = new_inv.id2path(file_id)
377
378
if specific_files:
379
if (not is_inside_any(specific_files, old_path)
380
and not is_inside_any(specific_files, new_path)):
381
continue
382
383
if kind == 'file':
384
old_sha1 = old_tree.get_file_sha1(file_id)
385
new_sha1 = new_tree.get_file_sha1(file_id)
386
text_modified = (old_sha1 != new_sha1)
387
else:
388
## mutter("no text to check for %r %r" % (file_id, kind))
389
text_modified = False
390
391
# TODO: Can possibly avoid calculating path strings if the
392
# two files are unchanged and their names and parents are
393
# the same and the parents are unchanged all the way up.
394
# May not be worthwhile.
395
396
if old_path != new_path:
397
delta.renamed.append((old_path, new_path, file_id, kind,
398
text_modified))
399
elif text_modified:
400
delta.modified.append((new_path, file_id, kind))
401
elif want_unchanged:
402
delta.unchanged.append((new_path, file_id, kind))
403
else:
404
kind = old_inv.get_file_kind(file_id)
405
old_path = old_inv.id2path(file_id)
406
if specific_files:
407
if not is_inside_any(specific_files, old_path):
408
continue
409
delta.removed.append((old_path, file_id, kind))
410
411
mutter('start looking for new files')
412
for file_id in new_inv:
413
if file_id in old_inv:
414
continue
415
new_path = new_inv.id2path(file_id)
416
if specific_files:
417
if not is_inside_any(specific_files, new_path):
418
continue
419
kind = new_inv.get_file_kind(file_id)
420
delta.added.append((new_path, file_id, kind))
421
422
delta.removed.sort()
423
delta.added.sort()
424
delta.renamed.sort()
425
delta.modified.sort()
426
delta.unchanged.sort()
427
428
return delta
Loggerhead is a web-based interface for Breezy
Version: 2.0.1